Power distribution network fault load transfer method and system generated based on candidate recovery channel

Abstract

The application provides a power distribution network fault load transfer method and system based on a candidate recovery channel, and relates to the technical field of power distribution network fault processing; the method comprises generating input features of a graph neural network, using the graph neural network to perform graph feature coding on a power distribution network after a fault, and generating a recovery value score; a compressed candidate recovery domain for a power loss area is generated according to the recovery value score; a load transfer mixed integer optimization model is constructed based on the compressed candidate recovery domain, the load transfer mixed integer optimization model is solved, and a fault recovery scheme of the power distribution network is obtained; the application can not only reduce the search space of the load transfer optimization problem under a complex fault scenario, but also can maintain the adaptability and feasibility of the recovery scheme to the operation constraints of the power distribution network, and improve the solving efficiency and engineering reliability of the fault self-healing decision.

Description

Technical Field

[0001] This invention relates to the field of distribution network fault handling technology, and in particular to a method and system for transferring faulty loads in distribution networks based on candidate recovery channels. Background Technology

[0002] With the continuous expansion of urban power distribution networks, a large number of distributed power sources, tie switches, and controllable loads are being connected. The load transfer problem after a power distribution network fault presents characteristics such as complex topology, numerous switch combinations, and diverse recovery paths. When a line or equipment fails, it is necessary to determine the operation of tie switches, the commissioning of distributed power sources, and load restoration schemes within a short period of time to restore the lost load as much as possible and ensure power supply to critical loads. However, traditional power distribution network fault recovery optimization methods typically require searching a large number of switch combinations and power supply paths across the entire network, which can easily lead to excessively large optimization models and long solution times, making it difficult to meet the actual needs of rapid decision-making in fault self-healing scenarios.

[0003] In recent years, mixed-integer optimization, heuristic algorithms, and reinforcement learning have been increasingly applied to the load transfer problem in distribution network fault scenarios. Mixed-integer optimization methods can effectively describe distribution network operational constraints and switch action constraints, but they suffer from heavy computational burdens under complex fault scenarios. Heuristic algorithms offer fast solution speeds, but the stability and constraint feasibility of recovery schemes are difficult to fully guarantee. End-to-end intelligent decision-making methods, while capable of learning fault recovery strategies, often lack interpretability of physical constraints and engineering reliability when directly outputting switch actions. Therefore, it is necessary to propose a method that combines the scene perception capabilities of graph neural networks with the constraint-solving capabilities of mathematical optimization models. By intelligently filtering and dynamically expanding candidate recovery channels after a fault, this approach can improve online solution efficiency under complex fault scenarios while ensuring the feasibility of load transfer schemes. Summary of the Invention

[0004] This invention provides a method and system for transferring fault loads in a distribution network based on candidate recovery channels, which addresses the shortcomings of existing technologies. It not only reduces the search space for load transfer optimization problems under complex fault scenarios, but also maintains the adaptability and feasibility of recovery schemes to distribution network operation constraints, thereby improving the solution efficiency and engineering reliability of fault self-healing decisions.

[0005] In a first aspect, the present invention provides a method for transferring faulty loads in a distribution network based on candidate recovery channels, comprising: Obtain the distribution network topology after the fault, and generate the operating status data of the distribution network under the fault scenario based on the distribution network topology. The input features of the graph neural network are generated based on the power distribution network topology and the operating status data. The input features are input into the graph neural network, and the graph neural network is used to encode the graph features of the distribution network after the fault, generating recovery value scores for lines, tie switches and recovery channels in the distribution network; Based on the recovery value score, a compressed candidate recovery domain is generated for the power loss area; Based on the compressed candidate recovery domain, a load transfer mixed integer optimization model is constructed, and the load transfer mixed integer optimization model is solved to obtain the fault recovery scheme of the distribution network.

[0006] Secondly, the present invention also provides a distribution network fault load transfer system based on candidate recovery channel generation, comprising: The data acquisition module is used to acquire the distribution network topology after a fault and generate the operating status data of the distribution network under the fault scenario based on the distribution network topology. The graph feature construction module is used to generate input features for the graph neural network based on the power distribution network topology and the operating status data. The recovery value assessment module is used to input the input features into the graph neural network, use the graph neural network to encode the graph features of the distribution network after the fault, and generate recovery value scores for the lines, tie switches and recovery channels in the distribution network. The candidate recovery channel generation module is used to generate compressed candidate recovery domains for the power loss area based on the recovery value score. The optimization solution module is used to construct a load transfer mixed integer optimization model based on the compressed candidate recovery domain, solve the load transfer mixed integer optimization model, and obtain the fault recovery scheme of the distribution network.

[0007] Thirdly, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the distribution network fault load transfer method based on candidate recovery channel generation as described above.

[0008] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the distribution network fault load transfer method based on candidate recovery channel generation as described above.

[0009] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the distribution network fault load transfer method based on candidate recovery channel generation as described above.

[0010] According to the specific embodiments provided in this application, the following technical effects are disclosed: The present invention provides a method and system for load transfer in a distribution network fault scenario based on candidate recovery channels. First, it generates operational status data of the distribution network under fault conditions based on the distribution network topology. Then, it generates input features for a graph neural network and inputs these features into the graph neural network. The graph neural network encodes the post-fault distribution network topology, load status, key load locations, tie switch status, distributed generation status, and line operation status using graph features, evaluating the recovery value of lines, tie switches, and candidate recovery channels, and generating recovery value scores for lines, tie switches, and recovery channels in the distribution network. Next, it uses a candidate recovery channel generation mechanism to filter high-value power supply paths and support channels based on recovery value, forming a compressed candidate recovery domain for the power outage area. Finally, under the constraints of candidate recovery channels, it constructs a mixed-integer optimization model for load transfer to solve for tie switch actions, distributed generation commissioning, and load recovery schemes. Through this collaborative architecture of graph learning and mathematical optimization, the present invention not only reduces the search space of the load transfer optimization problem under complex fault scenarios but also maintains the adaptability and feasibility of the recovery scheme to the distribution network operation constraints, improving the solution efficiency and engineering reliability of fault self-healing decisions. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0012] Figure 1 This is a schematic flowchart of the distribution network fault load transfer method based on candidate recovery channel generation provided by the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0014] In one exemplary embodiment, such as Figure 1As shown, this application provides a method for transferring fault loads in a distribution network based on candidate recovery channels. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method for transferring fault loads in a distribution network based on candidate recovery channels includes the following steps S101 to S105. Wherein: Step S101: Obtain the distribution network topology after the fault, and generate the operating status data of the distribution network under the fault scenario based on the distribution network topology; Step S102: Generate input features for a graph neural network based on the power distribution network topology and the operating status data; Step S103: Input the input features into the graph neural network, use the graph neural network to encode the graph features of the distribution network after the fault, and generate the recovery value scores of the lines, tie switches and recovery channels in the distribution network; Step S104: Generate a compressed candidate recovery domain for the power loss region based on the recovery value score; Step S105: Construct a load transfer mixed integer optimization model based on the compressed candidate recovery domain, solve the load transfer mixed integer optimization model, and obtain the fault recovery scheme of the distribution network.

[0015] By implementing steps S101 to S105 above, the distribution network fault load transfer method and system provided by the present invention based on candidate recovery channel generation first generates the operating status data of the distribution network under fault scenarios according to the distribution network topology. Then, it generates input features for a graph neural network and inputs the input features into the graph neural network. The graph neural network is used to encode the graph features of the distribution network topology, load status, key load locations, tie switch status, distributed power source status, and line operating status after the fault, evaluates the recovery value of lines, tie switches, and candidate recovery channels, and generates recovery value scores for lines, tie switches, and recovery channels in the distribution network. Next, the candidate recovery channel generation mechanism is used to screen high-value power supply paths and support channels according to the recovery value, forming a compressed candidate recovery domain for the power outage area. Finally, a mixed integer optimization model for load transfer is constructed under the constraints of candidate recovery channels to solve the tie switch action, distributed power source commissioning, and load recovery scheme. Through this collaborative architecture of graph learning and mathematical optimization, the present invention can not only reduce the search space of the load transfer optimization problem under complex fault scenarios, but also maintain the adaptability and feasibility of the recovery scheme to the distribution network operating constraints, and improve the solution efficiency and engineering reliability of fault self-healing decision-making.

[0016] As one embodiment of this application, step S101, which involves obtaining the distribution network topology after fault isolation and generating operating status data of the distribution network under fault scenarios based on the distribution network topology, includes steps S201 to S204. Wherein: Step S201: Obtain the basic topology data of the distribution network, which includes bus set, line set, switch set, tie switch set, load set, critical load set, distributed power source set and external power source node information; Step S202: Update the distribution network topology according to the status of the faulty line to obtain the distribution network topology after fault isolation; specifically: update the line commissioning status and switch isolation status according to the location of the faulty line or faulty equipment, thereby realizing the update of the distribution network topology and forming the distribution network topology after fault isolation.

[0017] Step S203: Perform connectivity analysis on the distribution network topology after fault isolation to obtain fault impact information. The fault impact information includes power-out load nodes, power-out load areas, power-available areas, network connectivity status after the fault, and fault impact range. Step S204: Perform power flow calculation based on the distribution network topology and fault impact information after fault isolation to obtain the operating status data, which includes node voltage, line load rate, line capacity utilization and distributed power source operating status.

[0018] Furthermore, in this embodiment of the application, in order to characterize the degree of load loss after a fault, a node is defined. The proportion of power outage load is: ; In the formula, Represents a node The proportion of power outage load; Represents a node Load power that was not restored after the fault; Represents a node The original load power. Understandably, the power loss load ratio is used to characterize the severity of the impact of a fault on different nodes.

[0019] Furthermore, in this embodiment of the application, in order to characterize the topological relationship between nodes and fault regions, the nodes are calculated. To the faulty line set Normalized distance: ; In the formula, Represents a node Normalized topological distance to the fault region; Represents a node With faulty line endpoints The shortest path distance between them; This represents the set of busbar nodes in the distribution network; it is understandable that the normalized distance is used to assist the graph neural network in identifying the scope of fault impact and potential recovery channels.

[0020] As one embodiment of this application, step S102, which involves generating input features for a graph neural network based on the distribution network topology and the operating status data, includes steps S301 to S302. Wherein: Step S301: Based on the distribution network topology, using the distribution network bus as the graph structure node and the lines, sectionalizing switch branches, and tie switch branches as the graph structure edges, construct the post-fault distribution network graph structure; the distribution network graph structure can be represented as: ; In the formula, Represents the set of nodes in a distribution network; This represents the set of edges formed by lines and switch branches; Step S302: Generate input features for a graph neural network based on the distribution network diagram structure and the operating status data; the input features include node features, edge features, tie switch features, and global fault scenario features. Wherein: For each node Its node characteristics can be represented as: ; In the formula, Indicates node voltage; Indicates whether the node is in operation; Indicates node load; Indicates the load at the node that has lost power; Indicates the critical load of a node; Indicates the distributed power capacity at the node; Indicates whether a node is an external power supply node; This represents the normalized distance from the node to the fault region; This indicates the encoding information of the region to which the node belongs.

[0021] For each edge Its edge characteristics can be represented as: ; In the formula, Indicates whether the line is in operation; Indicate whether the edge is a faulty edge; Indicate whether the edge is a handshake switch edge; Indicates the line resistance; Indicates the line reactance; Indicates line capacity; This indicates the line load rate.

[0022] For the tie switch side, further tie switch characteristics are constructed to describe its support capabilities in fault recovery, including the current state of the switch, the proportion of power loss loads on both sides, whether it is connected across regions, and the status of power supply resources on both sides.

[0023] By using the aforementioned node features, edge features, and tie switch features, the post-fault operation status of the distribution network can be transformed into a structured graph sample suitable for graph neural network processing.

[0024] It should also be noted that, in the embodiments of this application, the global fault scenario features can be represented as: ; In the formula, Indicates the number of faulty lines; Indicates the proportion of total power loss load in the system; Indicates the percentage of critical loads that have lost power; Indicates the number of connected components after a fault; This represents the proportion of available distributed power capacity. It should be noted that the global fault scenario characteristics are used to describe the overall severity of the fault scenario and the system's recoverability.

[0025] As one embodiment of this application, step S103, which involves using the graph neural network to encode the graph features of the faulted distribution network, includes steps S401 to S404. Wherein: Step S401: Map the node features, edge features, handshake switch features, and global fault scenario features to a unified latent space to obtain node embedding representation, edge feature embedding representation, handshake switch embedding representation, and global representation. Wherein: For nodes Its initial embedding is represented as: ; In the formula, Represents a node The initial embedding representation; and Indicates trainable parameters; Represents a node Node characteristics; This represents a non-linear activation function.

[0026] For the edge Its edge feature embedding is represented as: ; In the formula, Representing an edge Feature embedding representation; and Indicates trainable parameters; Representing an edge Edge features.

[0027] Step S402: Based on the node embedding representation and the edge feature embedding representation, a neighborhood information is aggregated using an edge-aware message passing mechanism to update the node embedding representation. Specifically, in step S402 of this embodiment, for the first node of the graph neural network... Layers, nodes Receive from neighboring nodes The message is: ; In the formula, Representing neighboring nodes Embedded Representation and edge embedding representation splicing; and Indicates message generation parameters; node The aggregated message is represented as: ; In the formula, Represents a node The set of neighboring nodes.

[0028] Based on the above aggregated message for the node The node state is updated using the following formula: ; In the formula, This represents the updated node embedding representation; and This represents the state update parameters; it can be understood that the multi-layer message passing achieved through the above update formula enables the graph neural network model to comprehensively perceive the fault location, power outage load distribution, interconnection switch connection relationship, and distributed power supply support capability.

[0029] Step S403: The node embedding representation, the handshake switch embedding representation, and the global representation are fused to obtain the recovery decision representation. The recovery decision representation can be expressed as: ; In the formula, This represents a graph-level representation obtained by embedding pooling all nodes; This represents the handshake switch support capability representation obtained by handshake switch feature pooling. This represents the global representation obtained by mapping global fault scenario features.

[0030] Step S404: Based on the recovery decision representation, calculate the recovery value score of the line, tie switch, and recovery channel. The calculation formula is as follows: ; In the formula, Indicates candidate recovery channel Recovery value score; Indicates candidate channel Its own feature embedding; and This represents the scoring parameter. The higher the score, the more suitable the channel is as a candidate recovery channel for load transfer in the current fault scenario.

[0031] In summary, by implementing the graph neural network encoding and scoring process in steps S401 to S404, the recovery value ranking of lines, tie switches, and candidate recovery channels can be obtained, providing a basis for the subsequent generation of candidate recovery channel sets and compression optimization solution.

[0032] As one embodiment of this application, step S104, which involves generating a compressed candidate recovery domain for the power loss region based on the recovery value score, includes steps S501 to S504. Wherein: Step S501: Based on the restoration value score, all restoration resources available for load transfer are sorted to form a restoration resource priority sequence; specifically, this application sorts all restoration resources available for load transfer based on the restoration value scores of lines, tie switches, and candidate restoration channels obtained in the preceding steps to form a restoration resource priority sequence; the restoration resource priority sequence can be expressed as: ; In the formula, This represents the set of candidate recovery resources sorted by recovery value score. This represents the complete set of candidate recovery resources, consisting of lines, tie switches, distributed power support paths, and load recovery paths. Indicates candidate recovery resources Recovery value score; This indicates an operation that sorts the data according to the rating from highest to lowest.

[0033] Step S502: For each power outage area or power outage load node, starting from available power supply nodes, external power supply nodes, or distributed power supply nodes, search for candidate recovery channels that can connect the power supply side and the power outage load side, and retain channels with higher recovery value scores according to the recovery resource priority sequence to form a candidate recovery channel set; the candidate recovery channel set can be represented as: ; In the formula, Indicates the node facing the power failure load. The set of candidate recovery channels; Indicates the connection between the power supply side and the power failure load node. The Candidate recovery channels Indicates the power outage load node The number of available candidate recovery channels.

[0034] Step S503: The candidate recovery channel set is filtered according to preset candidate channel constraints, and the candidate recovery channels corresponding to all power-loss load nodes are merged to form a compressed candidate recovery domain under the fault scenario, specifically including: First, based on the preset candidate channel width constraint, the candidate recovery channel set is filtered, retaining only the top-ranked channels. High-value candidate recovery channels: ; In the formula, This represents the set of candidate recovery channels after filtering. Indicates the power outage load node The maximum number of candidate recovery channels that can be retained; This indicates that the top performers were selected based on their restoration value scores. Operation of candidate channels; Secondly, all candidate recovery channels corresponding to the power failure load nodes are merged to form a compressed candidate recovery domain under the fault scenario: ; In the formula, This indicates a compressed candidate recovery domain; This represents the set of load nodes that lose power after a fault.

[0035] Step S504: When the compressed candidate recovery domain cannot meet the critical load recovery requirements or the optimization model is infeasible, spare recovery channels are gradually released from the spare recovery resource set according to the recovery value score to expand the compressed candidate recovery domain. The expanded compressed candidate recovery domain can be represented as: ; In the formula, This represents the expanded compression candidate recovery domain; This represents the set of backup recovery channels that were not initially selected; K represents the number of backup recovery channels released in each incremental release. This indicates the operation of selecting the top K backup recovery channels based on the recovery value score.

[0036] By implementing the candidate recovery channel screening and backup channel release mechanism in steps S501 to S503, this application can reduce the search space of the load transfer optimization problem without performing fixed feeder grouping, while retaining important recovery channels and key support resources, providing a foundation for subsequent compression optimization solutions.

[0037] As one embodiment of this application, step S105, which involves constructing a load transfer mixed-integer optimization model based on the compressed candidate recovery domain, includes steps S601 to S602. Wherein: Step S601: Establish load recovery variables, path selection variables, tie switch closing status variables, distributed power generation commissioning variables, and power output variables within the candidate recovery channel set.

[0038] Specifically, this application establishes a load recovery variable within the compressed candidate recovery domain. Path selection variables , Connecting switch closed state variable Distributed power generation commissioning variables and power output variables ;in: Load recovery variables Indicates load node Recovery power; path selection variables Indicates candidate recovery channel Whether it is selected; the status variable of the interconnection switch being closed. Indicates a connecting switch or line Closed state; distributed power generation commissioning variables Distributed power source Operational status; power output variables Indicates power supply . output power.

[0039] Step S602: Based on the load recovery variables, path selection variables, tie switch closing status variables, distributed power generation commissioning variables, and power output variables, construct a load transfer mixed integer optimization model including the optimization objective function and constraints; The optimization objective function aims to maximize load recovery, prioritize the recovery of critical loads, and control operational costs; the optimization objective function can be expressed as: ; In the formula, Represents the set of load nodes; Load Node The recovery weights are assigned to critical load nodes, with higher weights given to them. Indicates a controllable switch or circuit set; Indicates a switch or circuit The cost of the action; Represents a set of power sources; Indicates power supply The operating cost; This represents the weighting coefficients between different objective items.

[0040] The constraints include: load recovery constraints, candidate channel selection constraints, path and switch state coupling constraints, line capacity constraints, power output constraints, and supply-demand balance constraints, wherein: The load recovery constraint is used to ensure that the node recovery power does not exceed the original load demand. The expression for the load recovery constraint is: ; In the formula, Indicates load node The original load power.

[0041] The candidate channel selection constraint is used to ensure that each power-loss load node selects at most one candidate recovery channel. The expression for the candidate channel selection constraint is: ; In the formula, Indicates the power outage load node The number of candidate recovery channels.

[0042] The path-switch state coupling constraint is used to ensure that when a candidate recovery channel is selected, the relevant lines or tie switches on that channel should be in a closed or available state. The expression for the path-switch state coupling constraint is: ; In the formula, Indicates a line or switch Located in the candidate recovery channel superior.

[0043] The line capacity constraint is used to ensure that the recovery power flowing through the line does not exceed the line's allowable capacity. The line capacity constraint can be expressed as: ; In the formula, Represents a set of routes; Indicates the line The allowed capacity.

[0044] The power output constraint is used to ensure that the output power of external power sources and distributed power sources does not exceed their capacity limits. The power output constraint can be expressed as: ; In the formula, Indicates power supply Maximum output.

[0045] The supply-demand balance constraint is used to ensure that the power of the restored load within the candidate recovery domain is supported by available power sources. The supply-demand balance constraint can be expressed as: ; By implementing steps S601 to S602 above, this application forms a mixed integer optimization model for load transfer under candidate recovery channel constraints through the above-mentioned optimization objective function and constraints, and calls the optimization solver to solve it, thereby obtaining a fault recovery scheme that satisfies the candidate recovery domain and operational constraints. The fault recovery scheme includes tie switch operation, distributed power supply commissioning, and load recovery scheme.

[0046] As one embodiment of this application, after step S105, the method further includes outputting the distribution network fault load transfer result that meets the operational constraints and recovery requirements, including: Output the operating status of the interconnection switch and the load recovery result; Output the critical load recovery rate, the amount of unrecovered load, and the line operating status; Output the operational status and fault recovery indicators of distributed power sources; The generated fault recovery scheme will be used for fault self-healing and operation recovery of the distribution network.

[0047] Based on the same inventive concept, this application provides a distribution network fault load transfer system based on candidate recovery channels. The distribution network fault load transfer system based on candidate recovery channels provided by this invention is described below. The distribution network fault load transfer system based on candidate recovery channels described below can be referred to in correspondence with the distribution network fault load transfer method based on candidate recovery channels described above.

[0048] A distribution network fault load transfer system based on candidate recovery channel generation includes a data acquisition module, a graph feature construction module, a recovery value assessment module, a candidate recovery channel generation module, and an optimization solution module. Wherein: The data acquisition module is used to acquire the distribution network topology after a fault and generate the operating status data of the distribution network under the fault scenario based on the distribution network topology. The graph feature construction module is used to generate input features for the graph neural network based on the power distribution network topology and the operating status data. The recovery value assessment module is used to input the input features into the graph neural network, use the graph neural network to encode the graph features of the distribution network after the fault, and generate recovery value scores for the lines, tie switches and recovery channels in the distribution network. The candidate recovery channel generation module is used to generate compressed candidate recovery domains for the power loss area based on the recovery value score. The optimization solution module is used to construct a load transfer mixed integer optimization model based on the compressed candidate recovery domain, solve the load transfer mixed integer optimization model, and obtain the fault recovery scheme of the distribution network.

[0049] In one exemplary embodiment, this application also provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0050] In one exemplary embodiment, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0051] In one exemplary embodiment, this application provides a computer program product including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0052] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0053] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0054] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchain. The processors involved in the embodiments provided in this application may be, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc.

[0055] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0056] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for transferring faulty loads in a distribution network based on candidate recovery channels, characterized in that, include: Obtain the distribution network topology after the fault, and generate the operating status data of the distribution network under the fault scenario based on the distribution network topology. The input features of the graph neural network are generated based on the power distribution network topology and the operating status data. The input features are input into the graph neural network, and the graph neural network is used to encode the graph features of the distribution network after the fault, generating recovery value scores for lines, tie switches and recovery channels in the distribution network; Based on the recovery value score, a compressed candidate recovery domain is generated for the power loss area; Based on the compressed candidate recovery domain, a load transfer mixed integer optimization model is constructed, and the load transfer mixed integer optimization model is solved to obtain the fault recovery scheme of the distribution network.

2. The distribution network fault load transfer method based on candidate recovery channel generation according to claim 1, characterized in that, The process of obtaining the distribution network topology after fault isolation and generating operational status data of the distribution network under fault scenarios based on the distribution network topology includes: Acquire basic topology data of the distribution network, which includes bus sets, line sets, switch sets, tie switch sets, load sets, critical load sets, distributed power source sets, and external power source node information; The distribution network topology is updated based on the status of the faulty line to obtain the distribution network topology after fault isolation. Connectivity analysis is performed on the distribution network topology after fault isolation to obtain fault impact information, which includes power-out load nodes, power-out load areas, power-available areas, network connectivity status after the fault, and fault impact range. Based on the power flow calculation of the distribution network topology after fault isolation and the fault impact information, the operating status data is obtained. The operating status data includes node voltage, line load rate, line capacity utilization and distributed power source operating status.

3. The distribution network fault load transfer method based on candidate recovery channel generation according to claim 2, characterized in that, The step of generating input features for the graph neural network based on the distribution network topology and the operating status data includes: Based on the aforementioned distribution network topology, the distribution network bus is used as the graph structure node, and the lines, sectionalizing switch branches, and tie switch branches are used as the graph structure edges to construct the distribution network graph structure after a fault. The input features of the graph neural network are generated based on the power distribution network diagram structure and the operating status data; the input features include node features, edge features, tie switch features and global fault scenario features.

4. The distribution network fault load transfer method based on candidate recovery channel generation according to claim 3, characterized in that, The process of using the graph neural network to encode graph features of the post-fault distribution network includes: The node features, edge features, handshake switch features, and global fault scenario features are mapped to a unified latent space to obtain node embedding representation, edge feature embedding representation, handshake switch embedding representation, and global representation. Based on the node embedding representation and the edge feature embedding representation, an edge-aware message passing mechanism is used to aggregate neighborhood information in order to update the node embedding representation; The node embedding representation, the handshake switch embedding representation, and the global representation are fused to obtain the recovery decision representation; Based on the recovery decision characterization, the recovery value scores of the lines, tie switches, and recovery channels are calculated.

5. The distribution network fault load transfer method based on candidate recovery channel generation according to claim 1, characterized in that, The step of generating compressed candidate recovery domains for power loss regions based on the recovery value score includes: Based on the recovery value score, all recovery resources available for load transfer are sorted to form a recovery resource priority sequence; For each power outage area or power outage load node, search for candidate recovery channels that can connect the power supply side and the power outage load side, and retain channels with higher recovery value scores according to the recovery resource priority sequence to form a candidate recovery channel set; The candidate recovery channel set is filtered according to preset candidate channel constraints, and the candidate recovery channels corresponding to all power-loss load nodes are merged to form a compressed candidate recovery domain under the fault scenario; and, When the compressed candidate recovery domains cannot meet the preset recovery requirements, the backup recovery channels are released gradually according to the recovery value score.

6. The distribution network fault load transfer method based on candidate recovery channel generation according to claim 5, characterized in that, The construction of the load transfer mixed-integer optimization model based on the compressed candidate recovery domain includes: Within the candidate recovery channel set, establish load recovery variables, path selection variables, tie switch closing status variables, distributed power generation commissioning variables, and power output variables; Based on the load recovery variables, path selection variables, tie switch closing status variables, distributed power generation commissioning variables, and power output variables, a mixed integer optimization model for load transfer is constructed, including the optimization objective function and constraints. The optimization objective function aims to maximize load recovery, prioritize the recovery of critical loads, and control operational costs. The constraints include: load recovery constraints, path and switch state coupling constraints, line capacity constraints, power output constraints, and supply-demand balance constraints.

7. A distribution network fault load transfer system based on candidate recovery channel generation, characterized in that, include: The data acquisition module is used to acquire the distribution network topology after a fault and generate the operating status data of the distribution network under the fault scenario based on the distribution network topology. The graph feature construction module is used to generate input features for the graph neural network based on the power distribution network topology and the operating status data. The recovery value assessment module is used to input the input features into the graph neural network, use the graph neural network to encode the graph features of the distribution network after the fault, and generate recovery value scores for the lines, tie switches and recovery channels in the distribution network. The candidate recovery channel generation module is used to generate compressed candidate recovery domains for the power loss area based on the recovery value score. The optimization solution module is used to construct a load transfer mixed integer optimization model based on the compressed candidate recovery domain, solve the load transfer mixed integer optimization model, and obtain the fault recovery scheme of the distribution network.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the distribution network fault load transfer method based on candidate recovery channel generation as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the distribution network fault load transfer method based on candidate recovery channel generation as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the distribution network fault load transfer method based on candidate recovery channel generation as described in any one of claims 1 to 6.

Images